Weka (software)
| Company type | Open-source software |
|---|---|
| Industry | Machine learning, Data analysis |
| Founded | 1993 |
| Founder | University of Waikato |
| Headquarters | Hamilton , New Zealand |
Area served | Worldwide |
| Products | Machine learning and data analysis software |
| Owner | University of Waikato |
| Website | www |
Waikato Environment for Knowledge Analysis (Weka) is a collection of machine learning and data analysis open-source software distributed under the GNU General Public License. Originating from the University of Waikato in New Zealand, it serves as a companion to the publication "Data Mining: Practical Machine Learning Tools and Techniques".
Description
Weka provides a range of visualization utilities and algorithms designed for data analysis and predictive modeling, featuring graphical user interfaces to enhance usability. Its initial version, which was not Java-based, utilized a Tcl/Tk interface combined with modeling algorithms written in various languages and data preprocessing components developed in C, primarily aimed at analyzing agricultural datasets. Since 1997, the fully Java-implemented Weka 3 has broadened its scope to cover numerous fields, particularly in academic and research contexts.
Primary advantages for business users include:
- Cost-free distribution under the GNU General Public License.
- Platform independence enabled by Java technology.
- A broad array of data preprocessing and modeling methods.
- Intuitive graphical user interfaces.
Weka supports essential data mining operations such as data preprocessing, clustering, classification, regression, visualization, and feature selection. It accepts input data formatted in Attribute-Relational File Format (.arff). The software expects data arranged as a single flat file with fixed attributes, usually numeric or nominal types. Integration with SQL databases is facilitated through Java Database Connectivity, allowing it to handle outputs from database queries. Additionally, Weka offers access to deep learning capabilities via integration with Deeplearning4j.
Although Weka does not natively support multi-relational data mining, external utilities transform linked database tables into compatible formats for Weka processing. Its built-in algorithms do not cover sequence modeling.
Extension packages
Starting with version 3.7.2, Weka introduced a package manager that streamlines the installation of extension packages. Several functionalities formerly part of the core distribution were migrated to these extensions, promoting easier contributions and upkeep through a modular design that permits independent updates of both the core system and its extensions.
History
- The original development of Weka commenced in 1993 at the University of Waikato in New Zealand, initially constructed using Tcl/Tk, C, and makefiles.
- In 1997, Weka was completely rewritten in Java, incorporating implementations of modeling algorithms.
- Weka was honored with the SIGKDD Data Mining and Knowledge Discovery Service Award in 2005.
- In 2006, the Pentaho Corporation secured an exclusive license to utilize Weka for its business intelligence offerings, embedding it as the data mining and predictive analytics engine within its BI platform. After Pentaho’s acquisition by Hitachi Vantara, Weka now supports the PMI (Plugin for Machine Intelligence) open-source module.
Related tools
- Auto-WEKA streamlines machine learning processes within the Weka environment.
- Environment for DeveLoping KDD-Applications Supported by Index-Structures (ELKI) specializes in unsupervised learning and cluster analysis techniques.
- H2O.ai serves as an open-source platform for data science and machine learning applications.
- KNIME delivers machine learning and data mining solutions developed in Java.
- Massive Online Analysis (MOA), also created at the University of Waikato, facilitates mining of large-scale data streams.
- Neural Designer applies deep learning methodologies using C++ for advanced data mining.
- Orange is an open-source toolkit for data mining, machine learning, and visualization, built upon scikit-learn.
- RapidMiner offers a commercial Java-based machine learning platform that integrates Weka functionalities.
- scikit-learn is a popular Python library widely adopted for machine learning tasks.
See also
External links
- Official website hosted by the University of Waikato, New Zealand